BranchShine-CR：基于自条件CTC与一致性正则化的紧凑多语言国际音标转写
BranchShine-CR: Compact Multilingual IPA Transcription with Self-Conditioned CTC and Consistency Regularization
- International Centre for Neuromorphic Systems, Western Sydney University（西悉尼大学国际神经形态系统中心）
- Neurabuild
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
BranchShine-CR是一个25M参数的紧凑多语言IPA转写模型，通过自条件CTC和一致性正则化，在16,646条测试话语上达到4.47%字符错误率，较ZIPA-CTC-NS相对降低22.3%，参数少十二倍，适用于低资源设备端发音评估。
AI中文摘要：
我们提出BranchShine-CR，一个25M参数的模型，用于将多语言转写为国际音标（IPA）。它结合了对数梅尔特征、旋转位置E-Branchformer编码器、中间层自条件连接时序分类（CTC），以及跨增强视图的一致性正则化。在16,646条共享IPApack++测试话语上，它实现了4.47%的IPA字符错误率，相比ZIPA-CTC-NS相对降低了22.3%，参数数量约为其十二分之一，且从头开始训练。BranchShine-CR在所有41个数据集语言标签上也优于同等规模的NeMo Conformer基线。消融研究表明各组件在模型性能贡献中协同作用。这些发现支持在有限计算预算下实现紧凑的IPA识别能力，适用于低资源设备端发音评估应用。
英文摘要：
We introduce BranchShine-CR, a 25M-parameter model for multilingual transcription into the International Phonetic Alphabet (IPA). It combines log-mel features, a rotary-position E-Branchformer encoder, intermediate self-conditioned connectionist temporal classification (CTC), and consistency regularization across augmented views. On 16,646 shared IPApack++ test utterances, it achieves 4.47% IPA character error rate, a 22.3% relative reduction from ZIPA-CTC-NS, with approximately one-twelfth as many parameters while being trained from scratch. BranchShine-CR also outperforms a similarly sized NeMo Conformer baseline across all 41 dataset language labels. Ablation studies indicate the individual components synergetically acting in model performance contribution. These findings support compact IPA recognition capabilities under limited compute budget, for applications in low-resource on-device pronunciation assessment.
